In this study we first demonstrated that the CD

In this study, we first demonstrated that the CD44+ population isolated from the HCT-15 human colon cancer cell line not only exhibited multiple in vitro features of CSCs but also showed higher in vivo tumorigenicity. In accordance, knocking down CD44 expression in HCT-15 bosentan resulted in reduced soft agar colony-forming ability and the decreased expression of Bmi1, Snail and Twist1, whereas the increased expression of Snail, Twist1 as well as several ESC markers was found in CD44-overexpressing cells. We further showed that CD44 plays crucial roles in cancer initiation and stemness-maintenance in HCT-15 and HCT-116 cells which are modulated by the Snail-mediated suppression of a stemness inhibitor, miR-203. Finally, we further demonstrated that c-Src kinase, which is activated by hyaluronan (HA)/CD44 signaling, is responsible for Snail upregulation in these cells. Collectively, our results show for the first time that CD44 not only can be a single marker used for the isolation of CSCs from human colon cancer lines but also acts as a signal transducer leading to the suppression of miR-203. This suppression is crucial for the maintenance of stemness in these Co-CSCs.
Materials and methods
Results
Discussion CD133 has been used to identify various normal stem cells and CSCs. In fact, two recently published reports describe the isolation of colon cancer stem cells (Co-CSCs) using CD133 as a marker. The first publication showed that a small proportion (1 in 262) of CD133+ cells from primary colon cancers has a much higher tumorigenic potential in immunodeficient mice (Ricci-Vitiani et al., 2007). In addition to reporting a similar finding in vivo, the other study also demonstrated that CD133+ cells isolated from primary tumors formed spheres and could be maintained in an undifferentiated state in serum-free conditions (O\'Brien et al., 2007). Although CD133+ cells isolated from the HCT-15 human colon cancer line had a higher colony-forming ability than both the parental and CD133− cells (data not shown), they did not exhibit higher in vivo tumorigenicity after being cultivated in a defined medium for one month (Supplementary Fig. 2A and B). One possible explanation for this finding is that the purified CD133+ cells may gradually lose their stemness, and therefore their tumorigenicity, when cultured in a serum-free medium supplemented with EGF, bFGF and N2. Recent studies also demonstrated that Co-CSCs could be isolated using CD44 as a single selection marker (Chu et al., 2009; Du et al., 2008). On the other hand, an earlier work showed that the EpCAMhigh/CD44+/CD166+ cells are Co-CSCs (Dalerba et al., 2007). In support of the observations made by Chu and Du, we found that freshly isolated CD44+ HCT-15 cells exhibited a much higher soft agar colony-forming ability and tumor growth after xenograft than their CD44− counterparts (Fig. 1). These findings indicated that CD44 is a better marker than CD133 for selecting a stem cell-like population from HCT-15 cells. Although CD44 had been previously shown to participate in a wide variety of cellular functions including lymphocyte activation, recirculation and homing, hematopoiesis, and tumor metastasis (Misra et al., 2011), the functional aspects of this molecule in generating and/or maintaining CSCs have not yet been explored extensively. Accumulating evidence has shown that CD44 is not only a marker for Co-CSCs but also a critical factor for cancer initiation (Du et al., 2008). Interestingly, a recent study demonstrated that the induction of EMT stimulates cultured breast cells to adopt stem cell characteristics (Mani et al., 2008). Notably, our CD44 knockdown (Figs. 2A and B) and overexpression experiments (Figs. 2C and D) demonstrated that CD44 is important for maintaining the CSC properties of HCT-15 cells by regulating the EMT activators including Bmi1, Twist and Snail. However, the precise roles of Bmi1 and Twist in establishing and/or maintaining the stemness of CD44+ HCT-15 cells remain to be elucidated.